This article was accepted into the corpus but its outbound wikilinks were never NER-processed — typical at the deepest BFS hop or when the run's entity cap was reached. No expansion funnel to show.
| Yoshida (integrator) | |
|---|---|
| Name | Yoshida (integrator) |
| Caption | Symplectic composition schematic |
| Author | Hidehiko Yoshida |
| Introduced | 1990s |
| Field | Numerical analysis |
| Related | Suzuki–Yoshida composition, Runge–Kutta, symplectic integrator |
Yoshida (integrator) is a family of composition-based numerical integrators introduced to achieve higher-order accurate time propagation while preserving geometric structure for Hamiltonian systems. The method leverages compositions of lower-order maps to produce high-order symplectic schemes, combining ideas from splitting methods used in celestial mechanics, molecular dynamics, and Hamiltonian dynamics. Yoshida constructions are widely cited alongside algorithms such as the Verlet algorithm, Forest–Ruth integrator, and Suzuki fractal compositions.
Yoshida's work builds on earlier contributions by Hidehiko Yoshida, extending the lineage that includes Hermann E. M. Hermann-style operator splitting, G. M. Kreiss ideas, and innovations by L. Verlet, S. Blanes, F. Casas, M. Suzuki, and E. Forest. The approach composes second-order symmetric maps like the Störmer–Verlet or Leapfrog integrator into higher even-order schemes, analogous to constructions in Lie group integrators and Magnus expansion techniques. Yoshida compositions are often used in contexts involving the N-body problem, Molecular dynamics, Celestial mechanics, and long-term integrations where preservation of symplectic structure, energy behavior, and phase-space volume—properties emphasized by researchers like V. I. Arnold and J. Moser—is critical.
Given a separable Hamiltonian H = T(p) + V(q) typical of Hamiltonian mechanics formulations used in the N-body problem and Molecular dynamics, Yoshida constructs an order-2n integrator by composing symmetric second-order maps. Let S(h) denote a symmetric second-order map such as the Strang splitting or Störmer–Verlet step of timestep h. Yoshida defines coefficients a_i, b_i satisfying composition constraints and forms the map S_Y(h) = S(a_1 h) ◦ S(a_2 h) ◦ ... ◦ S(a_k h), with palindromic coefficient sequences to ensure time-reversibility akin to symmetric operator schemes. Coefficients are chosen to cancel lower-order error terms derived from the Baker–Campbell–Hausdorff expansion, following algebraic manipulations comparable to those used in Suzuki and Chin compositions. The minimal k for order 2n typically grows exponentially, reflecting combinatorial constraints observed in Butcher group and Runge–Kutta order theory.
Yoshida integrators are symplectic when built from symplectic second-order kernels, preserving the symplectic form and phase-space volume per Liouville-type theorems emphasized by Joseph Liouville and Poincaré. Time-symmetry (reversibility) is achieved by palindromic coefficient sets, connecting to results by Sanz-Serna and Hairer on geometric integration. Order conditions derive from eliminating terms in the Baker–Campbell–Hausdorff series up to a target degree, analogous to Butcher tableau constraints for Runge–Kutta methods. Stability properties relate to modified Hamiltonian theory and backward error analysis pioneered by Benettin and Giorgilli, ensuring near-conservation of invariants over exponentially long times for analytic systems as shown in studies by Skeel and Leimkuhler. However, Yoshida schemes can require negative substep coefficients for orders beyond four, a trade-off discussed in literature alongside alternatives by Blanes–Moan and Omelyan which seek optimized positive coefficients.
Extensions include Suzuki–Yoshida compositions that combine M. Suzuki fractal decompositions with Yoshida coefficient selection, and augmented schemes integrating force-gradient corrections from Chin to produce higher efficiency. Generalizations to nonseparable Hamiltonians involve splitting into more than two parts or embedding in Lie–Trotter frameworks used by Trotter and Kato. Multi-stage Yoshida-like constructions appear in symplectic partitioned methods, combining ideas from Partitioned Runge–Kutta and Gauss–Legendre implicit schemes, and adaptations exist for Lie group problems with references to Celledoni and Iserles. Optimization of coefficients for error constants has been pursued by Blanes, Casas, Sanz-Serna, and McLachlan to yield schemes tailored for Molecular dynamics thermostats like Nosé–Hoover.
Yoshida integrators are applied to long-term simulations in Celestial mechanics for planetary systems, to chaotic dynamics studies in the N-body problem, and to energy-preserving integration in Molecular dynamics for biomolecular modeling such as in AMBER and GROMACS workflows. They are also used in accelerator physics for tracking in particle accelerators, in plasma dynamics models studied by Horton and Birdsall, and in quantum semiclassical propagation where symplecticity and time-reversibility matter, paralleling work on Hagedorn wavepackets and the Wigner transform. Comparative studies benchmark Yoshida against Runge–Kutta–Nyström methods, symplectic implicit integrators like Gauss–Legendre, and optimized explicit splittings in simulation packages used by NASA and computational chemistry communities.
Practical implementation composes a reliable symmetric second-order kernel such as Velocity Verlet; for a fourth-order Yoshida integrator one uses a 3-stage composition with coefficients derived from algebraic cancellation, while higher orders use recursive formulas or precomputed coefficients from tables by Yoshida and Blanes–Moan. Pseudocode typically alternates kinetic and potential updates with timestep fractions a_i*h and b_i*h; careful handling of negative substeps and round-off is advised as discussed in works by Hairer, Leimkuhler, and Sanz-Serna. Example test problems include the pendulum and Kepler problem for validation, and larger benchmarks use the Solar System integrations and Lennard-Jones dynamics where conservation of energy and angular momentum over long times is measured. Implementations exist in libraries and software such as LSODE-style packages, molecular dynamics engines, and bespoke research codes contributed by computational groups in astrophysics and chemistry.
Category:Numerical integrators